OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

office-formula message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: Summary of 2010-02-16 office-formula teleconference


Summary of 2010-02-16 office-formula teleconference

Attendees:
David A. Wheeler
Andreas G.
Patrick D.
Eike R.
Hideki Hiura
Rob Weir
Dennis Hamilton
Eric Patterson


Discussion:
* Eike posted a new version 12 Feb 2010, but had a "hard time" -
  expects to upload a new version tonight (Tuesday).
* Patrick will work a few issues Wed-Thursday, upload Thursday night.
* Michael will process this on Thursday (remove change tracking, sections with tests etc.),
  posts on Friday.
  Michael needs to post it to the *office* TC, not the office-formula subcommittee.
* Propose to TC this Monday as a vote for a CD.
  Someone will probably want more time to review, so we'll probably start a 7-day electronic ballot at that point.

We had a long conversation about
http://tools.oasis-open.org/issues/browse/OFFICE-2380
Reference List (4.9 and 3.8)
Wheeler: A single Reference is also a ReferenceList (per 4.9), but in some cases
a multi-Reference ReferenceList should not be accepted by certain functions
and should be treated as an Error.
Patrick: I agree that there should be a "shall" that generates an Error... the problem is,
what is the condition that causes the error?
?: It's conceivable that an implementation could "do the right thing".
?: Some functions say "ReferenceList" but then require the ReferenceList to have only one reference.
Eike: Automatic conversion of Reference to ReferenceList would work.
Dennis: ReferenceList is both a syntactic and semantic notion (as well as Reference) - need to keep them
distinct.  The type is a semantic notion.
Wheeler: We need to resolve this for the final, not for the CD.
Eike has re-opened this comment.

Then we discussed:
http://tools.oasis-open.org/issues/browse/OFFICE-2345
(Error values.)
We need to clarify that the "/" is not ambiguous... once you see '#', maximally consume
the literal.  IE: "#N/A" is NA, not "#N" divided by "A".
Issue: Should Error be a new distinct type stored in OpenDocument?
Eric: This was problematic for Excel.  If you type '#N/A' that's a string;
when you round-trip, it's different.
Rob: There are legitimate reasons to create strings with #N/A in it.
Rob: We need the glue that describes how to store values in part 1... a little table of correspondances.
Should be a trivial mapping.

BIG ISSUE: Sounds like we should add a new "Error" type in OpenDocument part 1
to handle things completely.


Then there was a discusion about Complex, and if that had similar issues.


--- David A. Wheeler 


[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]